PI: Boon Thau Loo (University of Pennsylvania) Co-PI: William C. Regli (Drexel University) Senior Personnel: Joseph B. Kopena (Drexel University)
This research investigates scalable, knowledge-based middleware supporting content-based addressing and routing in mobile, networked systems. It incorporates and integrates two aspects: Ontological reasoning about system resources and declarative networking within routing components. At the application layer is OntoNet, a knowledge-based framework for representing and reasoning on system elements. Declarative, formal techniques provide service discovery and composition, content-based messaging, and distributed querying using OWL-Net, a subset of the OWL description logic. This work includes development of propagation strategies that are efficient and robust in mobile, networked environments. Network layer support is provided by declarative networks, a rule-based framework for compact, high-level protocol specifications. Declarative networking enables rapid prototyping and verification as well as online adaptation and meta-reasoning. This research will include extension of declarative networking to more readily support highly dynamic mobile wireless systems. The intellectual merit of this proposal is development of a unified, declarative framework for distributed organization of knowledge and information in real-world systems. It draws from many areas such as the Semantic Web, databases, and networking. Potential broader impacts are richer and more extensible platforms for real-world networks usable in emergency response, logistics, infrastructure monitoring, and ubiquitous computing.
This research investigates scalable, knowledge-based middleware supporting content-based addressing and routing in mobile, networked systems. It incorporates and integrates two aspects: Ontological reasoning about system resources and declarative networking within routing components. At the application layer is OntoNet, a knowledge-based framework for representing and reasoning on system elements. Declarative, formal techniques provide service discovery and composition, content-based messaging, and distributed querying using OWL-Net, a subset of the OWL description logic. This work includes development of propagation strategies that are efficient and robust in mobile, networked environments. Network layer support is provided by declarative networks, a rule-based framework for compact, high-level protocol specifications. Declarative networking enables rapid prototyping and verification as well as online adaptation and meta-reasoning. This project has lead to the following research outcomes: 1. PUMA: Policy-based Unified Multi-radio Architecture for Agile Mesh Networking. We use declarative networking techniques for configuring ad-hoc wireless networks used by OntoNet. The big innovation here is the introduction of formal tools based on constraint solvers to aid in the configuration process. Our proposed PUMA system is a declarative constraint-solving platform for policy-based routing and channel selection in multi-radio wireless mesh networks. In PUMA, users formulate channel selection policies as optimization goals and constraints that are concisely declared using the PawLog declarative language. To efficiently execute PawLog programs in a distributed setting, PUMA integrates a high performance constraint solver with a declarative networking engine. The PUMA system is available at http://netdb.cis.upenn.edu/puma for download. 2. Cologne Platform. Cologne is a declarative optimization platform that enables constraint optimization problems (COPs) to be declaratively specified and incrementally executed in distributed systems. Cologne integrates the RapidNet declarative networking engine with the Gecode constraint solver. Cologne uses the Colog declarative language that combines distributed Datalog used in declarative networking with language constructs for concisely specifying goals and constraints used in COPs. Using case studies based on cloud and wireless network optimizations, we demonstrate that Cologne (1) can flexibly support a wide range of policy-based optimizations in distributed systems, (2) results in orders of magnitude less code compared to imperative implementations, and (3) is highly efficient with low overhead and fast convergence times.Using case studies based on cloud and wireless network optimizations, we demonstrate that Cologne (1) can flexibly support a wide range of policy-based optimizations in distributed systems, (2) results in orders of magnitude less code compared to imperative implementations, and (3) is highly efficient with low overhead and fast convergence times. The Cologne system is available at http://netdb.cis.upenn.edu/cologne for download. The lead graduate student has graduated and is now a full-time research staff at AT&T Labs Research.
